In this post I explained how to couch an associative prediction that omitted predictions that were based solely on marginal statistics. The one catch was that if you asked for 5 items, you might not get 5, you could get 4 or 3 or even down to 0, since all predictions without a corresponding prediction node id were filtered out.
However, by leveraging the TopCount function, you can have your bread and butter it too (it’s late). Use syntax like the following:
(SELECT Product, $Probability FROM
PredictAssociation(Products, INCLUDE_STATISTICS, INCLUDE_NODE_ID)
WHERE $NodeID <> ”),
5) AS PredictedProducts
NATURAL PREDICTION JOIN